Operating plan · internal · 2026
The plan, gated.
How Mantle gets built and sold, in what order, and how we know it is working. The book argues the thesis; this plan runs it.
Model-shape. Pricing, projections, and unit economics stay as structure until the R3 willingness-to-pay research measures real demand. This plan commits to a sequence and a set of gates, not to numbers it has not earned. Everything is validation-gated: no phase begins until the prior gate clears.
Section 01
The bet, in one paragraph.
Mantle is the transaction workspace for top-producing listing agents, from accepted offer to closing. It does three things: tracks every deadline and contingency across a book of live deals, keeps a live closing-readiness grid per deal, and drafts the hard client messages in the agent's own voice. The agent adopts it alone; the compliance file it builds underneath becomes the key into the brokerage. The moat is two things no incumbent structurally has: a voice profile that compounds with every edited send, and a document layer classified by leak-damage by default. The full argument lives in the book.
The five decisions everything is downstream of
| Decision | Call |
|---|---|
| Buyer | Land the individual top producer (six or more deals under contract, the canonical quantifier across all docs); expand to the brokerage. |
| Wedge | The deal room: deadline and contingency tracking, closing-readiness, voice-profiled communication. |
| "Compliance" | Operational is the wedge (helps the agent close). Regulatory is the moat and the expand product (helps the broker). Custody is the third edge. |
| Voice profile | The one defensible asset. Bootstrapped from a sample plus imported comms; compounds from every edited send. |
| Capture | A secondary marketing-asset surface that also feeds the voice profile. |
Section 02
What winning looks like, by phase.
One outcome defines each phase. Everything else is diagnostic.
| Phase | Winning means |
|---|---|
| 0 · Validate | We know whether the wedge pain is real (R1) and whether the compliance file is acceptable (R4), cheaply, before building. |
| 1 · Wedge | 3 to 5 design-partner top producers run live deals through the deal room and stop keeping a backup checklist. |
| 2 · Adopt | Agents pay, adoption repeats beyond the founder's own network, and the voice profile is proven to compound. |
| 3 · Expand | One brokerage standardizes on the compliance file Mantle was already producing for free. |
Section 03
Operating principles.
Seven rules the plan does not break.
- Validation before build. R1 and R4 gate everything downstream. We do not write production code against an unproven wedge.
- Land the agent, expand the broker. A short sale to one person who feels the pain weekly beats a committee sale to a brokerage.
- Integrate, do not duplicate. Sit above the mandated paper tool (dotloop, SkySlope). Never ask the agent to switch off it.
- Custody-minimizing by default. Hold document status, not payload. Harden only the catastrophic handful, and only after R7.
- Voice is the moat. Every feature either feeds the voice profile or defends it. Features that do neither wait.
- Solo-adoptable or it is not the wedge. If an agent cannot say yes without the brokerage, it belongs in the expand product.
- Honest kill-criteria. Each phase has a gate that can end the project. Better to stop early than build on a wrong assumption.
Section 04
Roadmap: phased and gated.
Durations are relative, not calendar commitments. A phase ends when its exit gate clears, not when a clock runs out.
Answer R1, R4, and R5 cheaply, before writing production code.
Entry gateNone. Start here.
Key work
- 6 to 8 ride-alongs and interviews with top producers (R1)
- 5 to 10 brokerage compliance-officer and 1 to 2 E&O-underwriter conversations (R4)
- A voice-bootstrap technical spike on one agent (R5)
- Mandate-landscape mapping for target brokerages (R2); willingness-to-pay probing (R3); competitive teardown (R6)
Exit gate · all three must holdR1 shows closing-readiness is a top-three pain, and R4 shows the compliance file is acceptable to a compliance officer or E&O carrier, and R5 shows a voice profile trains from a sample plus sent mail. If any one fails, stop or pivot before building.
3 to 5 design partners using the deal room on live deals.
Entry gatePhase 0 exit gate cleared.
Key work
- Build the wedge MVP: tracking, the closing-readiness grid, voice drafting
- Integrate one paper tool per the R2 landscape
- Onboard design partners and instrument real usage
Exit gatePartners run live deals through Mantle without a backup legal pad; voice-draft acceptance rate holds; partners retain across multiple closes.
Agents pay, and adoption repeats without the founder in the room.
Entry gatePhase 1 exit gate cleared.
Key work
- Set pricing from R3; build founder-free onboarding
- Ship the document-sensitivity layer
- Run the secure-handoff feasibility spike (R7)
Exit gatePaid retention across closes; the voice profile demonstrably compounds (a late draft reads sharper than an early one); a growing stack of audit-ready closed files.
One brokerage standardizes on the compliance file.
Entry gatePhase 2 cleared, with audit-ready files stacked from agents inside a single brokerage.
Key work
- Build the broker compliance and admin tier
- Walk the closed-file stack into that brokerage's compliance office
- Sell per-seat or per-transaction
Exit gateOne brokerage rollout live; expand revenue turned on.
Section 05
Research as the execution backbone.
Not a side quest. It is the spine of Phase 0 and the gate on everything after. R1 and R4 are make-or-break.
| ID | Question it answers | Method | Ph. | Gates |
|---|---|---|---|---|
| R1 | Is closing-readiness a top-three pain or a nice-to-have? | 6 to 8 ride-alongs; map the real chase-and-close workflow. | 0 | The wedge itself. |
| R2 | Can an agent adopt Mantle without broker buy-in? | Determine each brokerage's mandated tool and what stacks on top. | 0 | Adoptability. |
| R3 | What will an agent pay, and on what model? | Probe current spend and model reaction inside R1. | 0 | Pricing. |
| R4 | Will compliance officers and E&O carriers accept an AI file? | 5 to 10 compliance interviews plus 1 to 2 E&O underwriters; show the mockup. | 0 | The expand motion. |
| R5 | Is a sample plus sent mail enough to train a usable voice? | Train a profile on one agent; blind-test against real writing. | 0 | The moat. |
| R6 | Where does the occupied category leave a wedge open? | Hands-on teardown of Folio and Open To Close. | 0 | Positioning. |
| R7 | Is the hardened secure-handoff buildable solo? | Legal and technical read on GLBA, SHIELD, and carrier requirements. | 2 | The custody build. |
Section 06
Product build order.
Build in gate order. Each layer earns the next. The sequenced version, with epics, gates, and dependencies, lives in the roadmap.
- Tracking. Every active deal on one screen, every deadline and contingency, sorted by what needs attention now.
- Closing-readiness grid. The live financing, documents, contingencies, and disclosures state per deal.
- Voice drafting. Client updates in the agent's own writing; the profile compounds from edits.
- Document-sensitivity layer. Every document classified by leak-damage; handling driven off that classification.
- Hardened secure-handoff. Replace email for the catastrophic few. Gated by R7; may need a partner.
What Mantle does not build. The paper of record and signatures stay in dotloop, SkySlope, Authentisign. The CRM and lead pipeline stay in Follow Up Boss, kvCORE. The brokerage back office stays in Lone Wolf, Brokermint. Mantle links and reads; it never duplicates.
Section 07
Go-to-market, one target at a time.
- Region. Westchester and lower Fairfield: dense with top producers, an attorney-state closing process, and the founder's direct reach into the market.
- Target brokerages. The top-producer-rich ones: Houlihan Lawrence, Compass, Sotheby's, Julia B. Fee, William Pitt.
- Founder access advantage. Direct, warm access to top producers in-region. This makes the land motion a short sale, not a cold one, and it is not easily copied by an out-of-market founder.
- Design-partner pipeline. Identify, ride-along (R1), pilot commitment, live use, reference. Each closed reference lowers the cost of the next.
- Sales cadence. Founder-led, one agent at a time. Lead with voice and closing-readiness, where incumbents are weakest. Hold compliance for the broker expand.
- Expand trigger. Once enough audit-ready closed files exist from agents inside one brokerage, approach that brokerage's compliance officer with the stack already on the desk.
Section 08
Business model, in shape.
The structure is committed; the figures wait on R3.
An agent subscription, expensed by the agent, is the base, with a per-transaction option for lower-volume agents. A brokerage tier sits on top for compliance oversight and multi-agent administration, sold per seat or per transaction. Revenue follows the product: it starts with the agent who adopts alone and expands to the brokerage that standardizes on the file Mantle was already producing.
Unit-economics drivers (structure, not numbers)
| Driver | What sets it |
|---|---|
| Price anchor | What the agent already spends on a transaction coordinator, dotloop, and AI tools. R3 measures it. |
| Acquisition cost | Low in-region: founder-led, then referral inside a tight professional network. |
| Retention | The voice switching cost plus the accumulating file. Both grow while the agent does nothing. |
| Expansion | Agents-per-brokerage. The land motion seeds the expand for free. |
Section 09
Team and hiring.
Hire against a cleared gate, not a hope.
- Now. Founder solo. The founder runs design-partner sales, product direction, and the Phase 0 research, and keeps build help lean.
- First hire, founding engineer. Gated to the Phase 0 exit gate plus committed design partners. Not before there is something proven to build.
- Second hire, design-partner success and ops. Gated to repeatable adoption in Phase 2, when onboarding must run without the founder.
- Compliance and legal counsel. Engaged as an advisor before any Tier-3 custody build (R7), not as a full-time hire.
Section 10
Cadence and metrics.
Weekly: design-partner touchpoints, deals tracked, blockers surfaced, voice-draft acceptance rate. Monthly: phase-gate review, research-track progress, pipeline health.
Leading indicators (structure, not targets)
The one number that matters, per phase: partners on live deals, then paid retention, then brokerage rollout.
Section 11
Risks and kill-criteria.
Each risk maps to the research that resolves it and to what we do if the answer is bad.
| Risk | Resolved by | If the answer is bad |
|---|---|---|
| Broker-mandated tool lock-in | R2 | The solo-adoption wedge dies. Stop, or pivot to a broker-first sale. |
| Regulatory rejection of the AI file | R4 | The expand motion dies. The wedge may stand alone; reassess scope. |
| Voice bootstrap fails | R5 | The moat is gone. Kill, or rebuild the moat around the document layer. |
| Parsing and event detection unreliable | Phase 1 build | The tracking layer weakens. Scope down to manual-assisted tracking. |
| Occupied category moves faster than we differentiate | R6, ongoing | If voice plus custody are not enough of a wedge, stop before scaling spend. |
Section 12
Funding, an operating decision.
- Default: bootstrap Phase 0. Raise nothing until there is signal. The validation work is cheap relative to the dilution an early raise would cost.
- After the Phase 0 gate and committed design partners: an optional small pre-seed to buy MVP speed and founder time. Best sources are real-estate operator angels (who also open brokerage doors), proptech pre-seed funds, and NAR's REACH accelerator, whose distribution into brokerages and MLSs outweighs its check.
- Milestone-gated, not calendar-gated. Raise from strength with signal in hand, or do not raise. A priced round is not credible until R1 and R4 are answered.
Section 13
The next 90 days.
July to September 2026. Six concrete moves.
- Line up 6 to 8 ride-alongs with top producers in-region (R1).
- Run the voice-bootstrap spike on one willing agent (R5).
- Book 3 to 5 conversations with brokerage compliance officers and one E&O underwriter (R4).
- Map the mandated-tool landscape for the five target brokerages (R2).
- Draft the wedge MVP spec: data model, screens, sprint plan.
- Convert at least 3 ride-alongs into design-partner commitments.
The gate to Phase 1 is items 1, 4, and 6 clearing with a green read on R1, R4, and R5. Everything after waits on that.